• We are sorry, but NCBI web applications do not support your browser and may not function properly. More information
Logo of nihpaAbout Author manuscriptsSubmit a manuscriptNIH Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
Arthritis Rheum. Author manuscript; available in PMC Jan 1, 2010.
Published in final edited form as:
PMCID: PMC2718690
NIHMSID: NIHMS75079

Systems Biology Analysis of Sjögren’s Syndrome and MALT Lymphoma Development in Parotid Glands

Abstract

OBJECTIVE

To identify key target genes and activated signal pathways associated with the disease pathogenesis by conducting a systems analysis of parotid gland manifesting primary Sjögren’s syndrome (pSS) and pSS/mucosa-associated lymphoid tissue (pSS/MALT) lymphoma phenotypes.

METHODS

A systems biologic approach was used to analyze parotid gland tissues obtained from non-pSS, pSS and pSS/MALT lymphoma patients. Concurrent expression microarray profiling and proteomic analysis were performed followed by weighted gene co-expression network analysis (WGCNA).

RESULTS

Gene co-expression modules related to pSS and pSS/MALT lymphoma are significantly enriched with genes known to be involved in immune/defense response, apoptosis, cell signaling, gene regulation, and oxidative stress. A detailed functional pathway analysis indicates that the pSS-associated modules are enriched with genes involved in proteasome degradation, apoptosis, signal peptides (MHC) class I, complement activation, cell growth and death, and integrin-mediated cell adhesion. The pSS/MALT-associated modules are enriched with genes involved in translation, ribosome, protease degradation, signal peptides (MHC) class I, G13 signaling pathway, complement activation, and Integrin-mediated cell adhesion. The combined analysis of gene expression and proteomics data implicates six highly connected hub genes for distinguishing pSS from non-pSS controls, and eight hub genes for distinguishing pSS/MALT lymphoma from pSS.

CONCLUSION

Systems biologic analysis of pSS and pSS/MALT parotid glands reveals pathways and molecular targets associated with the disease pathogenesis. The identified gene modules/pathways provide further insights into the molecular mechanisms of pSS and pSS/MALT lymphoma. The identified disease hub genes represent promising targets for therapeutic intervention, diagnosis, and prognosis.

INTRODUCTION

Sjögren’s syndrome (SS) is a systemic disease characterized by an autoimmune attack on salivary and lacrimal glands, resulting in dry eye and dry mouth symptoms. Without the lubricating and protecting function of saliva the oral cavity is highly susceptible to infections, including rampant caries and candidiasis, dysphagia, oral pain and discomfort. Similarly, the lack of tear lubrication results in irritation, foreign body sensation, blurring of vision and ocular surface disease (keratoconjunctivitis sicca). Histologically, SS is characterized by infiltration of exocrine glandular tissue by predominantly CD4+ T lymphocytes. The glandular epithelial cells express high levels of HLA-DR, leading to the speculation that these cells are presenting antigen to the invading T cells. There is also evidence of B cell activation with autoantibody production and an increase in B cell malignancy. Patients with pSS have significantly higher risk for lymphoma than the general population (1-3).

Understanding of the molecular mechanisms responsible for pSS and its progression to mucosal-associated lymphoid tissue lymphoma (pSS/MALT) requires molecular profiling analysis of the affected salivary gland tissues (e.g., parotid glands) from these patients. Two recent studies have investigated the global gene expression in minor salivary glands of pSS patients using microarray profiling. In the first study, a clear pattern of genes involved in chronic inflammation and apoptosis was demonstrated, including the chemokines, cytokines, MHC genes, lymphocyte activation factors, type I IFN genes and Bcl-2-like 2 (4). The second study indicated that IFNs and expression of 23 genes in the IFN pathways, including two Toll-like receptors (TLR8 and TLR9), were activated in minor salivary glands of SS patients compared with controls (5). These findings support the pathogenic interaction between the innate and the adaptive immune system in pSS.

Weighted gene co-expression network analysis (WGCNA) is a systems biologic analysis method that has been successfully used to identify disease pathways and their key constituents. It basically identifies gene co-expression modules based on unsupervised clustering of microarray data and explore both gene significance (differential expression) and connectivity for each gene (6, 7). WGCNA has been successfully used to identify key genes in glioblastoma multiforme (8), primate brain development (9) and inflammatory processes such as atherosclerosis (10, 11). In this study, we have performed transcriptomic analysis of pSS and pSS/MALT lymphoma in human parotid glands and identified gene co-expression modules using WGCNA. Functional pathway analysis of the genes in those modules was then conducted, and target genes were revealed and correlated with proteomic analysis.

MATERIALS AND METHODS

Patients and tissues

Human parotid gland (hPG) tissues from non-pSS control subjects (n=8) and patients with pSS (n=9) and pSS/MALT lymphoma (n=6) were obtained at the University Medical Center Groningen (UMCG) in the Netherlands with the Institutional IRB approval and patients’ consent. Clinical and histological features for the enrolled subjects are listed in Supplementary Table 1. All subjects enrolled in this study were female Caucasians.

All pSS and pSS/MALT patients underwent an incisional biopsy of one of the parotid glands under local anesthesia (4). The diagnosis of pSS was based on the US-EU criteria (12). pSS/MALT patients were based on a diagnosis of extranodal MALT lymphoma according to the World Health Organization classification (13) and a concomitant diagnosis of SS according to the US-EU criteria (12). Control parotid biopsies (non-pSS control) were obtained from patients with squamous cell carcinoma of the oral cavity or oropharynx without involvement of the parotid gland, and undergoing a neck dissection as part of the surgical treatment of their malignancy. These patients did not have subjective mouth or eye dryness, and no signs of lacrimal or salivary gland dysfunction. Parotid tissue was removed from the dorsal caudal lobe during a neck dissection procedure.

Next to the snap-frozen parotid gland biopsies used for the systems analysis, part of the parotid tissues were fixed in 4% phosphate buffered formalin, embedded in paraffin blocks, and cut at a thickness of 3 μm for histopathological examination. The slides were stained with hematoxylin and eosin. Histopathological evaluation of the biopsies from non-pSS subjects revealed a normal architecture of the tissue, while in biopsies from pSS and pSS/MALT subjects showed alterations matching Sjögren’s syndrome (focus score > 1 and presence of lymphoepithelial lesions) (14). In addition, specimens suspected for pSS/MALT were subjected to immunohistochemical analysis for CD3, CD20, CD10, BCL6 and cytoplasmic immunoglobulins (kappa, lambda, IgM, IgG and IgA). The classification described by De Vita et al. was used for distinction between reactive benign lymphoproliferation and MALT lymphoma (15).

mRNA and protein sample preparation

The protein and RNA isolation system (PARIS™ kit, Ambion, Austin, TX) was used to extract mRNAs and proteins from snap-frozen hPG tissues ranging from 10 mg to 30 mg. Total RNA isolation was performed using the RNeasy kit supplied with the PARIS kit. The total lysate was centrifuged for 3 min at 10,000×g before adding equal volume of 100% ethanol. The mixture was then washed with the provided buffer and RNA was eluded with 50μl elution buffer. The resulted total RNA was subjected to RNase-free DNase (DNase I, Ambion) treatment followed by ethanol precipitation. Finally the total RNA was dissolved into 15 μl DNase/RNase-free H2O and quantified with the Nanodrop spectrophotometer (ND-1000, Nanodrop Technology, Wilmington, DE).

The total protein amount of each tissue samples was measured using the 2-D Quant Kit (GE Healthcare, Piscataway, NJ). Each sample was then purified with 2-D Cleanup Kit (GE Healthcare), and re-dissolved in rehydration buffer. Due to the limited amount of the hPG tissues, equal amount of tissue samples from the controls, pSS or pSS/MALT patients were pooled, respectively, for subsequent 2-DE analysis.

Gene-expression profiling

The RiboAmp® RNA Amplification system (MDS Analytical Technologies, Mountain view, CA) was used to perform one round of linear amplification of 2.5 μg total mRNAs from each hPG tissue sample. The synthesized cDNAs were in vitro transcribed to aRNAs and then biotinylated using the GeneChip Expression 3’-Amplification Reagents for in vitro transcription labeling (Affymetrix, Santa Clara, CA). The labeled aRNAs (15μg each sample) were then fragmented and quality assessed with the 2100 Bioanalyzer (Agilent Technologies, Palo Alto, CA).

Gene-expression profiling was performed with the Affymetrix Human Genome U133 Plus 2.0 arrays, which contains > 54,000 probe sets representing > 47,000 transcripts and variants. Fragmented cRNAs were hybridized overnight to the microarrays. After high-stringency wash to remove the unbound probes, the hybridized chips were stained and scanned according to the Affymetrix standard expression protocol. The acquired images were processed with the Affymetrix MicroArray robust multi-array average (RMA) dChip software (16), and subsequently imported into the statistical software R for computation analysis. We excluded two control and one pSS/MALT arrays based on the following quality control criteria: mean signal intensity > 50, present call > 45%, array outlier < 5% and single outliers < 1%.

2-DE analysis

In total, 300-μg proteins from each pooled samples (non-SS control, pSS and pSS/MALT) were used for 2-DE analysis. Isoelectric focusing was performed with 17-cm immobilized pH gradient (IPG) gel strip (pI 3-10 NL, Bio-Rad, Hercules, CA), and SDS-PAGE was performed with 8-16% pre-cast gradient gels (Bio-Rad). Fluorescent Sypro-Ruby stain (Invitrogen, CA) was used for protein detection, and the resulting gel images were analyzed using PDQuest (Bio-Rad).

Protein identification

Proteins spots were excised using a spot-cutting robot (Proteome Works, Bio-Rad). Each gel spot was treated with DTT and iodoacetamide and finally digested with 10 ng trypsin at 37° for overnight. The resulting peptides were analyzed using nano-LC (Eksigent Technology) with linear ion trap MS (LTQ XL, Thermo-Fisher Scientific). The LC separation was performed with a PepMap C18 column (75μm×150mm; particle size 3μm, Dionex, Sunnyvale, CA). The acquired MS/MS data were searched against the human IPI (International Protein Index) database using SEQUEST (Thermon-Fisher Scientific).

Weighted gene co-expression network analysis (WGCNA)

The details about WGCNA can be found at www.genetics.ucla.edu/labs/horvath/Co-expressionNetwork. We constructed gene co-expression networks by soft thresholding the Pearson correlation matrix between the gene expression profiles as described previously (17). Because microarray data can be noisy and the number of samples is often small, therefore it is useful to emphasize strong correlations and punish weak correlations. A natural way of doing this is to define the connection strength between 2 genes as a power ß >1 of the correlation coefficient. The correlation between two gene expression profiles determines the connection strength (adjacency) between two genes in the network. By adding up the connection strengths for each gene, a single number (called connectivity) is produced to describe how strongly that gene is connected (correlated) to all other genes in the network. The next step in network construction is to identify groups of genes with similar patterns of connection strengths by searching for genes with high “topological overlap” (6, 7). Gene co-expression modules were formed by hierarchical clustering without regard to the disease status (unsupervised clustering). Our custom-made R software function for WGCNA and the data are available upon request. A glossary of WGCNA terminology can be found in Supplementary Table 2, and a workflow for WGCNA and correlation with proteomic analysis can be found in Supplementary Figure 1.

Statistical analysis

We used Student’s t-test to test for differential expression. Apart from the p-value, we also used the q-value function in the R software to compute a q-value for each gene. The q-value measures the proportion of positives incurred, or local false discovery rate, when the corresponding t-test p-value is called significant.

Using terminology from WGCNA, we refer to the absolute value of the t-test as the gene significance. Genes with gene significance larger than 2 (|t|>2) are significantly differentially expressed at a 0.05 level. We defined a measure of module significance by averaging the gene significance measures for each module. The expression profiles of the genes inside a given module are represented by the module eigengene, that is, the first principal component of the scaled gene expression profiles. Analogous to the case of a single gene, we define the module eigengene significance of differential expression between two groups by the absolute value of the corresponding t-test statistic.

Module Membership

Module detection involves certain parameter choices. For some genes, it may be difficult to decide whether they belong to a particular module or whether they belong to more than one module. Instead of a binary indicator of module membership (MM), it can be advantageous to use a fuzzy measure of MM. We define a measure of MM as the correlation between a gene expression profile and the module eigengene, which is the most representative gene, in a specific module. Thus, the closer MM is to 1 or -1, the stronger the evidence that this gene is part of the module (6, 7).

Gene ontology and functional pathway analysis

Gene ontology analysis was performed using DAVID database and EASE software. We also used Biocarta and the Kyoto Encyclopaedia of Genes and Genomes (KEGG; http://www.genome.jp/kegg/) to study whether the disease related co-expression modules are significantly enriched with known molecular interactions and reaction networks.

Results

The purpose of this study is to conduct concurrent transcriptomic and proteomic analysis of parotid gland tissues from patients with pSS and pSS/MALT in order to understand potential key target genes and signal pathways underlying the disease mechanism. We applied weighed gene co-expression network analysis (WGCNA) to analyze a gene-expression data set comprised of non-pSS control, pSS and pSS/MALT samples. Proteomic data were also combined with the WGCNA analysis to reveal candidate genes concordantly altered at mRNA/protein levels.

WGCNA - pSS versus non-pSS control

Modules are defined as branches of a hierarchical cluster tree. Eight different modules were identified from WGCNA analysis of gene-expression data between pSS and non-pSS control groups (Fig. 1A). We then examined whether any of these modules were associated with the disease status. Toward this end, we defined a measure of gene significance by the absolute value of the t-test statistic for testing differential expression between patients and controls. We observed that the Turquoise, Red and Brown modules were enriched with differentially expressed genes between pSS and control groups (module gene significance score > 2) (Fig. 1B). The module significance (mean gene significance) for the Turquoise, Brown and Red modules corresponded to the following p-values: p=0.009, p=0.02 and p=0.03, respectively, indicating that these modules (and their respective module eigengenes) are strongly associated with the pSS in parotid glands. The heatmap plot shown in Fig. 1C suggests that the Turquoise module genes are highly correlated and some of the representative “hub” genes are presented in Fig. 1F. Most of the genes in Turquoise module are also significant differentially expressed, including the 22 most significant genes (GS>3) that can well segregate pSS from control groups (Fig. 1D).

Figure 1Figure 1
Module and hub gene detection based on gene expression analysis of pSS patients and controls. (A) Hierarchical cluster tree used for module detection. Modules correspond to branches of the tree and are assigned the same color as indicated by the color-band ...

Gene ontology (GO) analysis indicated three disease-related modules (Turquoise, Red and Brown) are significantly enriched with genes in the following biological processes: immune and defense response (humoral immune response, antigen presentation and processing, response to pathogens, stress and external stimulus), antigen presentation/processing, apoptosis and cell death, intracellular signaling and cell communication, and RNA metabolism, processing and splicing (Supplementary Table 3). We also performed a post-hoc analysis in order to identify whether specific pathways were represented in these co-expression modules. The pathways corresponding to these modules with p<0.05 are listed in Supplementary Table 4. The most significant pathway is the proteasome degradation (p=0.0003). Other significant pathways include the apoptosis (p=0.002), signal peptides (MHC) class I, complement activation (p=0.002), cell communication (p=0.002), cell growth and death (p=0.002), and integrin-mediated cell adhesion (p=0.002). Note that these p values were the unadjusted p values without considering the multiple comparison problem. We also performed pathway analysis based on the differentially expressed genes (fold change >2 and p<0.05) and very similar signal pathways were identified (data not shown), suggesting that the strong biological signal can also be found using alternative statistical procedures. An advantage of WGCNA is that it naturally allows one to prioritize genes within disease related modules based on their connectivity to other disease related genes.

Highly connected intramodular “hub’’ genes play prominent role in maintaining the module structure. As reviewed in the glossary (Supplementary Table 2), we define the connectivity for each gene based on its Pearson correlation with all of the other genes in the module (6). It turns out that the intramodular connectivity is also highly related to the absolute value of the module membership measure (correlation>0.95). To screen for disease related candidate genes, we use a screening criterion based on module membership to disease-related modules and gene significance (differential expression). Fig. 1E presents the scatter plot of gene significance versus intramodular connectivity for the genes in the “Turquoise” module. One particularly promising intramodular “hub” gene in the Turquoise module, apoptotic peptidase activating factor 1 (APAF1), is shown in Fig. 1F. APAF1 is a cytoplasmic protein that binds to cytlosolic cytochrome C in the presence of ATP or dATP during apoptosis, leading to self-oligomerization followed by association with, and activation of, caspase-9 (18). E2F1 transcription factor can regulate the expression of APAF1, which directly link the deregulation of the pRB pathway with apoptosis. Independently of the pRB pathway, APAF-1 is also a direct transcriptional target of p53, suggesting that p53 might sensitize cells to apoptosis by increasing APAF1 levels (19, 20). APAF1 has been previously implicated in autoimmune diseases and chemoresistance of B-cell lymphoma (21, 22).

WGCNA - pSS versus pSS/MALT

The gene co-expression network analysis of pSS and pSS/MALT samples (i.e. control samples were omitted) identified 8 gene co-expression modules (Fig. 2A), including five modules (Turquoise, Red, Yellow, Brown, Green) comprised of highly differentially expressed genes (average T-test > 2, Fig. 2B). The module significance for Turquoise, Red, Yellow, Brown, Green modules corresponded to the following p-values: p=0.0001, p=0.0006, p=0.002, p=0.003 and p=0.03, respectively.

Figure 2
Modules identified from gene expression analysis of pSS versus pSS/MALT lymphoma. (A) Hierarchical cluster tree used for module detection. These modules are different from those found in our analysis of pSS versus controls. (B) Disease-related modules ...

The heatmap plot of the Turquoise module genes (Fig. 2C) illustrates that the genes are highly differentially expressed and are highly correlated. Scatter plots of gene significance versus intramodular connectivity are presented in Fig. 2D-E. While there is a significant correlation between intramodular connectivity and gene significance, we find it useful to use both module membership (intramodular connectivity) and gene significance as complementary variables to screen for disease progression related genes that are centrally located inside disease related modules. We performed GO and functional pathway analysis of the genes from the five pSS/MALT-associated modules (Supplementary Table 5 & 6). The most significant pathways across all the five modules are translation, ribosome and protease degradation pathways. Other interesting pathways include signal peptides (MHC) class I, CD47-IAP with avB3, G13 signaling pathway, structure of caps and SMACs, complement activation, cell communication, and integrin-mediated cell adhesion.

Integration of proteomics and microarray data

Using 2-DE/MS, we identified a number of proteins associated with pSS and pSS/MALT lymphoma. Among the 115 proteins showing above 3-fold elevated levels, 20 proteins were up-regulated in pSS compared to non-pSS control and pSS/MALT lymphoma subjects, 25 proteins were up-regulated in both pSS and pSS/MALT lymphoma as compared to non-pSS control, and 70 proteins were up-regulated in pSS/MALT lymphoma compared to both non-pSS control and pSS. We found that 45% of the up-regulated proteins had an mRNA transcript (gene expression level) that was concordantly differentially expressed. Supplementary Tables 7 & 8 list the genes concordantly altered at mRNA/protein levels between non-pSS control and pSS or between pSS and pSS/MALT.

Candidate key target genes selection for pSS and pSS/MALT pathogenesis

A panel of candidate genes based on the Turquoise, Brown and Red modules for distinguishing pSS from non-pSS control subjects are listed in Table 1. These genes showed significant differential expression as indicated by the p-values, and they were an important component of the pSS-related modules as reflected by the strongly positive or negative values of MMTurquoise, MMBrown and MMRed. Similarly, a list of candidate genes for distinguishing pSS/MALT lymphoma from pSS, as implicated by both WGCNA and proteome analysis, is listed in Table 2. These candidate targets were selected from supplementary tables 7 & 8, respectively.

Table 1
Six candidate genes for distinguishing pSS from non-pSS controls. These 6 genes were implicated by proteomics data and WGCNA (from Supplementary Table 7). The T-test statistic (and its p-value) measures differential expression. MMTurquoise, MMBrown, and ...
Table 2
Eight candidate genes for distinguishing pSS/MALT from pSS. These 8 genes were implicated by proteome analysis and WGCNA (from Supplementary Table 8). See Table 1 for other details.

Discussion

Elucidating the molecular mechanisms underlying pSS and pSS/MALT lymphoma remains an important challenge. As a result of the dearth of molecular markers for pSS, the diagnosis of this devastating disease often lags behind disease onset by almost a decade. The objective diagnostic criterion of lymphocytic infiltration requires an invasive biopsy procedure, though it is currently a routine procedure in the diagnostic work-up for SS (3). The gain in knowledge regarding the cellular mechanisms of T and B lymphocyte activity in pathogenesis of pSS has resulted in novel opportunities (e.g., anti-TNF-alpha, IFN- alpha, anti-CD20, anti-CD22 and anti-B cell-activating factor) for therapeutic intervention, however none of these agents have been approved yet for treatment of pSS and pSS/MALT (23-26). Understanding the molecular basis of pSS and pSS/MALT is critical for improving target-based therapies and developing diagnostic and prognostic criteria (e.g., lymphoma development). We show that a systems-level analysis of high-throughput gene expression and proteomics data can implicate disease-related pathways and their key constituents.

In this study, we first identified gene co-expression modules related to pSS or pSS/MALT based on analysis of gene expression data with WGCNA. Using this method, we then determined the intramodular connectivity for each gene, which measures how connected, or coexpressed, a given gene is with respect to the genes of a particular module (6). This provided highly connected intramodular hub genes that are most significant for maintaining the structure and function of the disease modules. This was followed by conducting GO and functional pathway analysis to understand the most significant biological processes and pathways associated with the disease modules. Finally, a panel of candidate key target genes for pSS and pSS/MALT pathogenesis, as implicated by both WGCNA and proteomic analysis are presented for future biological validation.

GO analysis indicated that the pSS-associated modules (Turquoise, Brown and Red) are significantly enriched with genes known to be involved in immune/defense response, antigen presentation/processing, apoptosis and cell death pathways. This observation is consistent with our previous study, which revealed the salivary mRNAs associated with apoptosis and immune responses in pSS patients (27). The most significant GO terms (biological processes) were well preserved between pSS/MALT-associated modules (Turquoise, Red, Yellow, Brown and Green) and pSS-associated modules (Turquoise, Red and Bronw), although they showed dissimilar significances. The fact that unsupervised clustering based on a co-expression measure resulted in modules enriched for biologically important processes suggests that these modules are a robust feature of the molecular architecture of pSS and pSS/MALT.

Proteasome degradation appears to be the most significant pathway associated with the pSS modules. Since the proteasome system has a pivotal role in the control of the immune response, it might be involved in the pathogenesis of autoimmune disorders such as pSS. In fact, recent studies have shown that the expression of proteasome subunits LMP2 and LMP7 were significantly altered in pSS (28, 29). Functional pathway analysis also suggested that apoptosis of epithelial cells in parotid glands significantly contributes to the pathogenesis of pSS. Apoptotic cell death may be induced by cytotoxic T cells through the release of perforin and granzymes. Cleavage of certain autoantigens during apoptosis such as caspase-mediated proteolysis of alpha-fodrin and poly(ADP-ribose) polymerase (PARP) may reveal immunocryptic epitopes that could potentially induce autoimmune response and lead to tissue destruction on the development of pSS (30, 31). Apoptosis can also be induced by the interaction of Fas ligand (FasL/CD95L), expressed by T lymphocytes, with Fas (Apo-1/CD95) on epithelial cells. Fas ligand (FasL), and its receptor Fas are essential in the homeostasis of the peripheral immune system (32). All of these previously reported genes, including CD40, Fas, FasL, perforin, caspases and PARP, were found significantly altered in pSS in our study, which further confirmed the role of apoptotic pathways for impaired function of salivary glandular cells in pSS. In pSS/MALT-associated modules, two most significant pathways are translation and ribosome pathways. However, most of the other pathways (e.g., proteasome degradation pathway) were highly preserved between pSS and pSS/MALT modules. Considering the limited sample size and heterogeneous cell population in the disease patients’ parotid gland biopsies, further studies are required to confirm these findings regarding key target genes and activated signal pathways. Since MALT lymphoma only constitutes about half the lymphomas, it may also be interesting to look into those discovered genes between lymphoma with pSS and without pSS.

The proteins elevated in pSS patients are functionally related to immune/defense response, apoptosis, cell-cell adhesion and anti-oxidative stress whereas many of the proteins up-regulated in pSS/MALT lymphoma are related to signal transduction, gene regulation, apoptosis, immune response, and oxidative stress. Some of these targets have been linked to lymphoma previously, and two cancer-related proteins, Rho-GDP dissociation inhibitor (Rho-GDI) and cyclophilin A (CypA), are particularly of biological significance. The over-expression of Rho-GDI has been observed in most of the human tumors and may associate with the development of drug resistance in breast and lymphoma cancer cells (33). In fact, Rho-GDI is an anti-apoptotic molecule that mediates cellular resistance to chemotherapy agents. The mechanism for the anti-apoptotic activity of Rho-GDI may derive from its ability to inhibit caspase-mediated cleavage of Rac1 GTPase. CypA has been reported to be over-expressed in cancer cells, especially in solid tumors. Over-expression of CypA prevented hypoxia- and chemo-induced apoptosis, and this was associated with the suppression of reactive oxygen species generation and depolarization of mitochondrial membrane potential (34). Both Rho-GDI and CypA were highly connected genes based on WGCNA, and they showed concordant over-expression in pSS/MALT lymphoma than in pSS and non-pSS controls based on proteomic and transcriptomic analyses.

The systems approach led to the discovery and identification of a number of candidate genes that are bears the potential to be key molecular targets for pSS and pSS/MALT pathogenesis in parotid glands. The fact that integrated WGCNA with proteome analysis identified the known disease-related genes provides a proof-of-principle of the systems biologic strategy in studying pSS and its progression. These candidate genes, previously linked or unlinked to pSS or pSS/MALT lymphoma, can now be tested for their biological involvement by developing models for in vitro and in vivo testing. This process will be iterative (model building and testing) and will eventually lead to the identification of key molecular targets of pSS and pSS/MALT pathogenesis.

In summary, this paper provides proof-of-concept data that a systems analysis of pSS and pSS/MALT lymphoma using tools of transcriptomics, proteomics and WGCNA have led to the identification of distinct biological pathways and key target candidate genes related to pSS pathogenesis and lymphoma progression. We found consistent disease-related features with respect to molecular networks involved in various aspects of proteasome degradation, immune/defense response, apoptosis, cell signaling, gene regulation, and oxidative stress. It has become clear in the past decade that the initiation of autoimmunity is a multifactorial and complex process that requires genetic components to synergize with multi-etiological events (35). With the large scale gene expression and proteomic profiling and the use of systems biologic methods such as WGCNA to integrate multi-dimensional data, we have narrowed down the co-expression modules and associated gene determinants based on their interaction, concordance, and connectivity. The synergetic role of the identified candidate key target genes in modules/pathways can now be further validated with the use of animal or cell models, and will help elucidate the molecular mechanisms. The discovered candidate genes, when validated, can be translated into early diagnosis/prognosis biomarkers and molecular targets for therapeutic intervention.

Supplementary Material

Supp Data

Supp Tab 1

Supp Tab 2

Supp Tab 3

Supp Tab 4

Supp Tab 5

Supp Tab 6

Supp Tab 7

Supp Tab 8

Acknowledgments

This work was supported by the PHS grant R01-DE17593 (David T. Wong). Shen Hu was supported by the PHS grants R21-CA122806 and R03-DE017144.

References

1. Fox RI. Sjögren’s syndrome. Lancet. 2005;366(9482):321–31. [PubMed]
2. Hansen A, Lipsky PE, Dorner T. B-cell lymphoproliferation in chronic inflammatory rheumatic diseases. Nat Clin Pract Rheumatol. 2007;3(10):561–9. [PubMed]
3. Ryu OH, Atkinson JC, Hoehn GT, Illei GG, Hart TC. Identification of parotid salivary biomarkers in Sjogren’s syndrome by surface-enhanced laser desorption/ionization time-of-flight mass spectrometry and two-dimensional difference gel electrophoresis. Rheumatol (Oxford) 2006;45(9):1077–86. [PubMed]
4. Gottenberg JE, Cagnard N, Lucchesi C, Letourneur F, Mistou S, Lazure T, Jacques S, Ba N, Ittah M, Lepajolec C, Labetoulle M, Ardizzone M, Sibilia J, Fournier C, Chiocchia G, Mariette X. Activation of IFN pathways and plasmacytoid dendritic cell recruitment in target organs of primary Sjögren’s syndrome. Proc Natl Acad Sci U S A. 2006;103(8):2770–5. [PMC free article] [PubMed]
5. Hjelmervik TO, Petersen K, Jonassen I, Jonsson R, Bolstad AI. Gene expression profiling of minor salivary glands clearly distinguishes primary Sjögren’s syndrome patients from healthy control subjects. Arthritis Rheum. 2005;52(5):1534–44. [PubMed]
6. Zhang B, Horvath S. A general framework for weighted gene co-expression network analysis. Stat Appl Genet Mol Biol. 2005;4:17. [PubMed]
7. Ghazalpour A, Doss S, Zhang B, Wang S, Plaisier C, Castellanos R, Brozell A, Schadt EE, Drake TA, Lusis AJ, Horvath S. Integrating genetic and network analysis to characterize genes related to mouse weight. PLoS Genet. 2006;2(8):e130. [PMC free article] [PubMed]
8. Horvath S, Zhang B, Carlson M, Lu KV, Zhu S, Felciano RM, Laurance MF, Zhao W, Qi S, Chen Z, Lee Y, Scheck AC, Liau LM, Wu H, Geschwind DH, Febbo PG, Kornblum HI, Cloughesy TF, Nelson SF, Mischel PS. Analysis of oncogenic signaling networks in glioblastoma identifies ASPM as a molecular target. Proc Natl Acad Sci U S A. 2006;103(46):17402–7. [PMC free article] [PubMed]
9. Oldham MC, Horvath S, Geschwind DH. Conservation and evolution of gene coexpression networks in human and chimpanzee brains. Proc Natl Acad Sci U S A. 2006;103(47):17973–8. [PMC free article] [PubMed]
10. Gargalovic PS, Imura M, Zhang B, Gharavi NM, Clark MJ, Pagnon J, Yang WP, He A, Truong A, Patel S, Nelson SF, Horvath S, Berliner JA, Kirchgessner TG, Lusis AJ. Identification of inflammatory gene modules based on variations of human endothelial cell responses to oxidized lipids. Proc Natl Acad Sci U S A. 2006;103(34):12741–6. [PMC free article] [PubMed]
11. Gong KW, Zhao W, Li N, Barajas B, Kleinman M, Sioutas C, Horvath S, Lusis AJ, Nel A, Araujo JA. Air-pollutant chemicals and oxidized lipids exhibit genome-wide synergistic effects on endothelial cells. Genome Biol. 2007;8(7):R149. [PMC free article] [PubMed]
12. Vitali C, Bombardieri S, Jonsson R, Moutsopoulos HM, Alexander EL, Carsons SE, et al. Classification criteria for Sjogren’s syndrome: a revised version of the European criteria proposed by the American-European Consensus Group. Ann Rheum Dis. 2002;61(6):554–558. [PMC free article] [PubMed]
13. Jaffe ES, Harris NL, Stein H, Vardiman JW. World Health Organization classification of tumours: pathology and genetics of tumours of haematopoietic and lymphoid tissues. Lyon, France: IARC Press; 2001.
14. Pijpe J, Kalk WWI, van der Wal JE, Vissink A, Kluin PM, Roodenburg JLN, et al. Parotid gland biopsy compared with labial biopsy in the diagnosis of patients with primary Sjogren’s syndrome. Rheumatology. 2007;46(2):335–341. [PubMed]
15. De Vita S, De Marchi G, Sacco S, Gremese E, Fabris M, Ferraccioli G. Preliminary classification of nonmalignant B cell proliferation in Sjögren’s syndrome: perspectives on pathobiology and treatment based on an integrated clinico-pathologic and molecular study approach. Blood Cells Mol Dis. 2001;27(4):757–66. [PubMed]
16. Irizarry RA, Bolstad BM, Collin F, Cope LM, Hobbs B, Speed TP. Summaries of Affymetrix GeneChip probe level data. Nucleic Acids Res. 2003;31(4):e15. [PMC free article] [PubMed]
17. Zhang B, Horvath S. A general framework for weighted gene co-expression network analysis. Stat Appl Genet Mol Biol. 2005;4(1):Article17. [PubMed]
18. Li P, Nijhawan D, Budihardjo I, Srinivasula SM, Ahmad M, Alnemri ES, Wang X. Cytochrome c and dATP-Dependent Formation of Apaf-1/Caspase-9 Complex Initiates an Apoptotic Protease Cascade. Cell. 1997;91(4):479–489. [PubMed]
19. Robles AI, Bemmels NA, Foraker AB, Harris CC. APAF-1 Is a Transcriptional Target of p53 in DNA Damage-induced Apoptosis. Cancer Res. 2001;61(18):6660–6664. [PubMed]
20. Moroni MC, Hickman ES, Lazzerini Denchi E, Caprara G, Colli E, Cecconi F, Muler H, Helin K. Apaf-1 is a transcriptional target for E2F and p53. Nat Cell Biol. 2001;3(6):552–8. [PubMed]
21. Ting JP, Kastner DL, Hoffman HM. CATERPILLERs, pyrin and hereditary immunological disorders. Nat Rev Immunol. 2006;6(3):183–195. [PubMed]
22. Sun Y, Orrenius S, Pervaiz S, Fadeel B. Plasma membrane sequestration of apoptotic protease-activating factor-1 in human B-lymphoma cells: a novel mechanism of chemoresistance. Blood. 2005;105(10):4070–4077. [PubMed]
23. Meijer JM, Pijpe J, Bootsma H, Vissink A, Kallenberg CG. The Future of Biologic Agents in the Treatment of Sjögren’s Syndrome. Clin Rev Allergy Immunol. 2007;32(3):292–7. [PMC free article] [PubMed]
24. Pijpe J, van Imhoff GW, Spijkervet FKL, Roodenburg JLN, Wolbink GJ, Mansour K, Vissink A, Kallenberg CGM, Bootsma H. Rituximab treatment in patients with primary Sjögren’s syndrome: An open-label phase II study. Arthritis Rheum. 2005;52(9):2740–2750. [PubMed]
25. Pers JO, Devauchelle V, Daridon C, Bendaoud B, Le Berre R, Bordron A, Hutin P, Renaudineau Y, Dueymes M, Loisel S, Berthou C, Saraux A, Youinou P. BAFF-modulated repopulation of B lymphocytes in the blood and salivary glands of RITUXIMAB-TREATED patients with Sjögren’s syndrome. Arthritis Rheum. 2007;56(5):1464–1477. [PubMed]
26. Sankar V, Brennan MT, Kok MR, Leakan RA, Smith JA, Manny J, et al. Etanercept in Sjogren’s syndrome: a twelve-week randomized, double-blind, placebo-controlled pilot clinical trial. Arthritis Rheum. 2004;50(7):2240–5. [PubMed]
27. Hu S, W J, Meijer J, Ieong S, Xie Y, Yu T, Zhou H, Henry S, Vissink A, Pijpe J, Kallenberg C, Elashoff D, Loo JA, Wong DT. Salivary proteomic and genomic biomarkers for primary Sjögren’s syndrome. Arthritis Rheum. 2007;56(11):3588–3600. [PMC free article] [PubMed]
28. Krause S, Kuckelkorn U, Dorner T, Burmester GR, Feist E, Kloetzel PM. Immunoproteasome subunit LMP2 expression is deregulated in Sjögren’s syndrome but not in other autoimmune disorders. Ann Rheum Dis. 2006;65(8):1021–1027. [PMC free article] [PubMed]
29. Egerer T, Martinez-Gamboa L, Dankof A, Stuhlmüller B, Dörner T, Krenn V, Egerer K, Rudolph PE, Burmester GR, Feist E. Tissue-specific up-regulation of the proteasome subunit beta5i (LMP7) in Sjögren’s syndrome. Arthritis Rheum. 2006;54(5):1501–1508. [PubMed]
30. Hayashi Y, Arakaki R, Ishimaru N. The role of caspase cascade on the development of primary Sjögren’s syndrome. J Med Invest. 2003;50(1-2):32–8. [PubMed]
31. Jimenez F, Aiba-Masago S, Al Hashimi I, Vela-Roch N, Fernandes G, Yeh CK, Talal N, Dang H. Activated caspase 3 and cleaved poly(ADP-ribose)polymerase in salivary epithelium suggest a pathogenetic mechanism for Sjogren’s syndrome. Rheumatol. 2002;41(3):338–342. [PubMed]
32. Bolstad AI, Eiken HG, Rosenlund B, Alarc-Riquelme ME, Jonsson R. Increased salivary gland tissue expression of Fas, Fas ligand, cytotoxic T lymphocyte-associated antigen 4, and programmed cell death 1 in primary Sj An external file that holds a picture, illustration, etc.
Object name is nihms75079ig1.jpg ren’s syndrome. Arthritis Rheum. 2003;48(1):174–185. [PubMed]
33. Zhang B, Zhang Y, Dagher MC, Shacter E. Rho GDP Dissociation Inhibitor Protects Cancer Cells against Drug-Induced Apoptosis. Cancer Res. 2005;65(14):6054–6062. [PubMed]
34. Choi KJ, Piao YJ, Lim MJ, Kim JH, Ha J, Choe W, Kim SS. Overexpressed cyclophilin A in cancer cells renders resistance to hypoxia- and cisplatin-induced cell death. Cancer Res. 2007;67(8):3654–62. [PubMed]
35. Blank M, Barzilai O, Shoenfeld Y. Molecular mimicry and auto-immunity. Clin Rev Allergy Immunol. 2007;32(1):111–8. [PubMed]
PubReader format: click here to try

Formats:

Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...

Links

  • MedGen
    MedGen
    Related information in MedGen
  • PubMed
    PubMed
    PubMed citations for these articles

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...